修正邮件外域多地址连接预算并保留失败原因

backend/utils/smtp/context_test.go

@@ -2,11 +2,70 @@ package smtp
 
 import (
 	"context"
+	"errors"
 	"net"
+	"net/netip"
 	"testing"
 	"time"
 )
 
+func TestDialResolvedAddressesFallsBackAfterTimeout(t *testing.T) {
+	client, server := net.Pipe()
+	defer client.Close()
+	defer server.Close()
+	addresses := []netip.Addr{netip.MustParseAddr("192.0.2.1"), netip.MustParseAddr("2001:db8::1")}
+	calls := 0
+	conn, err := dialResolvedAddresses(context.Background(), addresses, "25", 10*time.Millisecond, func(ctx context.Context, network, address string) (net.Conn, error) {
+		calls++
+		if network != "tcp" || address != net.JoinHostPort(addresses[calls-1].String(), "25") {
+			t.Fatalf("unexpected dial target: %s %s", network, address)
+		}
+		if calls == 1 {
+			<-ctx.Done()
+			return nil, ctx.Err()
+		}
+		if ctx.Err() != nil {
+			t.Fatal("fallback inherited exhausted attempt context")
+		}
+		return client, nil
+	})
+	if err != nil || conn != client || calls != 2 {
+		t.Fatalf("fallback failed: conn=%v calls=%d err=%v", conn, calls, err)
+	}
+}
+
+func TestDialResolvedAddressesHonorsCancellation(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	addresses := []netip.Addr{netip.MustParseAddr("192.0.2.1"), netip.MustParseAddr("192.0.2.2")}
+	calls := 0
+	_, err := dialResolvedAddresses(ctx, addresses, "25", time.Second, func(context.Context, string, string) (net.Conn, error) {
+		calls++
+		cancel()
+		return nil, context.Canceled
+	})
+	if calls != 1 || !errors.Is(err, context.Canceled) {
+		t.Fatalf("canceled parent was ignored: calls=%d err=%v", calls, err)
+	}
+}
+
+func TestDialResolvedAddressesRetainsFailures(t *testing.T) {
+	first := errors.New("first address failed")
+	second := errors.New("second address failed")
+	addresses := []netip.Addr{netip.MustParseAddr("192.0.2.1"), netip.MustParseAddr("192.0.2.2")}
+	calls := 0
+	_, err := dialResolvedAddresses(context.Background(), addresses, "25", time.Second, func(context.Context, string, string) (net.Conn, error) {
+		calls++
+		if calls == 1 {
+			return nil, first
+		}
+		return nil, second
+	})
+	if calls != 2 || !errors.Is(err, first) || !errors.Is(err, second) {
+		t.Fatalf("connection failures were lost: calls=%d err=%v", calls, err)
+	}
+}
+
 func TestSMTPContextCancelsGreeting(t *testing.T) {
 	client, server := net.Pipe()
 	defer server.Close()

backend/utils/smtp/smtp.go

@@ -62,7 +62,8 @@ func IsDeliveryUncertain(err error) bool {
 }
 
 const (
-	outboundSMTPDialTimeout         = 2 * time.Second
+	outboundSMTPDialTimeout         = 10 * time.Second
+	outboundSMTPConnectTimeout      = 30 * time.Second
 	outboundSMTPGreetingTimeout     = 5 * time.Second
 	outboundSMTPTLSHandshakeTimeout = 2 * time.Second
 	outboundSMTPTransactionTimeout  = 2 * time.Minute
@@ -215,9 +216,11 @@ func dialTCPContext(parent context.Context, addr string, publicOnly bool) (net.C
 		return nil, errors.New("smtp: invalid remote address")
 	}
 
-	ctx, cancel := context.WithTimeout(parent, outboundSMTPDialTimeout)
+	ctx, cancel := context.WithTimeout(parent, outboundSMTPConnectTimeout)
 	defer cancel()
-	addresses, err := resolveRemoteAddresses(ctx, host)
+	resolveCtx, resolveCancel := context.WithTimeout(ctx, outboundSMTPDialTimeout)
+	addresses, err := resolveRemoteAddresses(resolveCtx, host)
+	resolveCancel()
 	if err != nil {
 		return nil, err
 	}
@@ -228,18 +231,30 @@ func dialTCPContext(parent context.Context, addr string, publicOnly bool) (net.C
 	}
 
 	dialer := &net.Dialer{Timeout: outboundSMTPDialTimeout, KeepAlive: 30 * time.Second}
-	var lastErr error
+	return dialResolvedAddresses(ctx, addresses, port, outboundSMTPDialTimeout, dialer.DialContext)
+}
+
+// 为各地址分配独立连接预算，同时遵守父级总时限，避免首个地址耗尽全部回退时间。
+func dialResolvedAddresses(ctx context.Context, addresses []netip.Addr, port string, timeout time.Duration, dial func(context.Context, string, string) (net.Conn, error)) (net.Conn, error) {
+	var failures []error
 	for _, address := range addresses {
-		conn, err := dialer.DialContext(ctx, "tcp", net.JoinHostPort(address.String(), port))
+		if err := ctx.Err(); err != nil {
+			failures = append(failures, err)
+			break
+		}
+		attemptCtx, cancel := context.WithTimeout(ctx, timeout)
+		conn, err := dial(attemptCtx, "tcp", net.JoinHostPort(address.String(), port))
+		cancel()
 		if err == nil {
 			return conn, nil
 		}
-		lastErr = err
+		failures = append(failures, err)
 	}
-	if lastErr == nil {
+	if len(failures) == 0 {
 		return nil, errors.New("smtp: remote address is unavailable")
 	}
-	return nil, fmt.Errorf("smtp: connect to remote address: %w", lastErr)
+	// 保留各地址的失败原因，不能把最后一个未成功连接的地址当成唯一故障。
+	return nil, fmt.Errorf("smtp: connect to remote address: %w", errors.Join(failures...))
 }
 
 func resolveRemoteAddresses(ctx context.Context, host string) ([]netip.Addr, error) {